Skip to main content
. 2019 Oct 27;20(21):5343. doi: 10.3390/ijms20215343

Table 1.

Summary of the datasets generated for simulation experiments.

Dataset TV Average Cytosine Sites/Sample Samples/Group * Cut-Point Accuracy Sen. Spe. FDR
1. Model building & cross-validation 0.0356 1,000,000 3 0.3500 1.0000 1.0000 1.0000 0.0000
2. Model building & cross-validation 0.1332 1,000,000 3 0.9404 0.9998 0.9997 1.0000 0.0000
3. Model building & cross-validation 0.1845 1,000,000 3 0.9251 1.0000 1.0000 1.0000 0.0000
4. External data for validation 0.0356 1,000,000 50 0.3500 1.0000 1.0000 1.0000 0.0000
5. External data for validation 0.1332 1,000,000 50 0.9404 0.9998 1.0000 1.0000 0.0000
6. External data for validation 0.1845 1,000,000 50 0.9251 1.0000 0.9999 1.0000 0.0000
7. Model building & cross-validation 0.0356 1,000,000 50 0.3500 1.0000 1.0000 1.0000 0.0000
8. Model building & cross-validation 0.1332 1,000,000 50 0.8667 1.0000 1.0000 1.0000 0.0000
9. Model building & cross-validation 0.1845 1,000,000 50 0.8306 1.0000 1.0000 1.0000 0.0000
10. External data for validation 0.0356 1,000,000 50 0.3500 1.0000 1.0000 1.0000 0.0000
11. External data for validation 0.1332 1,000,000 50 0.8667 1.0000 1.0000 1.0000 0.0000
12. External data for validation 0.1845 1,000,000 50 0.8306 1.0000 1.0000 1.0000 0.0000

* Four independent control samples were used to build the reference sample (the centroid) for each simulation 1 to 6; while twenty were used for simulations 7 to 12. All the R scripts for these simulations are available at https://git.psu.edu/genomath/MethylIT_examples.